2.1 Investigation of the reaction of PdBr₂(PPh₃)₂ with CO in the absence and presence of PPh₃ (2 equiv)

The complex PdBr₂(PPh₃)₂ (2 mM) in THF (containing nBu₄NBF₄, 0.3 M) was characterized by a single overall bielectronic reduction peak R₁ at E^p_R1 = –0.875 V/SCE (gold disk electrode, i.d. 0.5 mm, scan rate: 0.2 V s^–1). An oxidation peak O₁ located at +0.100 V/SCE was observed on the reverse scan (Fig. 1a, full line). We have reported that the electrochemical reduction of PdBr₂(PPh₃)₂ generates an anionic complex [Pd⁰(PPh₃)₂Br]^– oxidized at O₁ (Eqs. (4) and (5)) [13].

In the presence of excess PPh₃ (2 equiv), the reduction peak R₁ was not significantly modified, whereas the oxidation peak current intensity of the electrogenerated Pd⁰ complex located at O₁ was slightly higher than in absence of PPh₃, due to a stabilization of the electrogenerated Pd⁰ complex by PPh₃ (Eq. (6)) [13].

Carbon monoxide was then bubbled into the cell to investigate any reaction of PdBr₂(PPh₃)₂ with CO – complexation [12] or reduction. In the presence of CO, the reduction peak R₁ of PdBr₂(PPh₃)₂ was slightly shifted to more negative potential (E^p_R^′1 = –0.895 V) with a lower peak current (Fig. 1a, dashed voltammogram). A new reduction peak R₂ of small intensity appeared at more negative potential, E^p_R2 = –1.025 V (Fig. 1a, dashed voltammogram). This suggests that, in the presence of CO, bromide ions were released in higher amount in the reduction of PdBr₂(PPh₃)₂ (Eq. (4)) than in the absence of CO. As reported, Br^– ions react with PdBr₂(PPh₃)₂ and expel one PPh₃ to generate an anionic Pd^II complex [Pd^IIBr₃(PPh₃)]^– (Eq. (7)) [14] less easily reduced at R₂.

The potential shift observed from R₁ to R′₁ for PdBr₂(PPh₃)₂ is a consequence of its involvement in the equilibrium in Eq. (7) (CE mechanism).

The complex PdBr₂(PPh₃)₂ was thus rather stable in THF in the presence of bubbling CO, because the overall reduction current of PdBr₂(PPh₃)₂ did not decrease with time (the current intensity of R′₁ being proportional to PdBr₂(PPh₃)₂ concentration). Consequently, at room temperature and atmospheric CO pressure, complexation of PdBr₂(PPh₃)₂ by CO did not occur. It is known that CO can reduce Pd^II salts to Pd⁰ [15] but when ligated by phosphine, as in PdBr₂(PPh₃)₂, the reduction by CO (under atmospheric pressure) did not occur in the time scale investigated here (1 h). The reduction of PdCl₂(PPh₃)₂ by CO does not occur even at high CO pressure [12]. No oxidation peak was detected when the scan was performed to oxidation first.

One notices that in the presence of CO, the oxidation peak O₁ of [Pd⁰(PPh₃)₂Br]^–, generated by the reduction of PdBr₂(PPh₃)₂, disappeared and two new oxidation peaks O₂ (E^p_O2 = +0.365 V) and O₃ (E^p_O3 = +0.595 V) were detected (Fig. 1a, dashed voltammogram). The latter was assigned to the oxidation of bromide ions by comparison to an authentic sample of nBu₄NBr. This confirms that, in the presence of CO, the electrochemical reduction of PdBr₂(PPh₃)₂ did not generate the anionic [Pd⁰(PPh₃)₂Br]^– complex but a new Pd⁰ complex, assigned to Pd⁰(PPh₃)₂(CO) (Eq. (8)), which was oxidized at O₂. However, its small oxidation peak current indicates that Pd⁰(PPh₃)₂(CO) was not very stable within the time scale of the cyclic voltammetry, due to the lack of stabilizing ligand or to insufficient CO pressure (Eq. (9), S = solvent).

The existence of the equilibrium in Eq. (9) has been evidenced by the investigation of the kinetics of oxidative addition of PhI with Pd⁰(PPh₃)₄ in the presence of CO (retarding effect of CO) [16].

In the presence of PPh₃ (2 equiv) and CO, the voltammogram of PdBr₂(PPh₃)₂ exhibited the only reduction peak R₁ (Fig. 1b), because the equilibrium in Eq. (7) was shifted towards its left hand-side due to the excess PPh₃. On the reverse scan, a single oxidation peak was detected at O₄ (E^p_O4 = +0.245 V), whose peak current was much higher than that of O₂ (compare Fig. 1a and b).

To characterize the Pd⁰ complex that was oxidized at O₄, a set of experiments was performed with an authentic sample of Pd⁰(PPh₃)₄, which contains 4 equiv PPh₃ per Pd⁰ as when PdBr₂(PPh₃)₂ was reduced in the presence of 2 equiv PPh₃. In THF, Pd⁰(PPh₃)₄ dissociates to Pd⁰(PPh₃)₃, SPd⁰(PPh₃)₃ and SPd⁰(PPh₃)₂ (Scheme 1, S = solvent) [17].

This is why the voltammogram of Pd⁰(PPh₃)₄ (2 mM) in THF (containing nBu₄NBF₄, 0.3 M) exhibited two close oxidation peaks O′₁ (E^p_O^′1 = +0.245 V) and O'₂ (E^p_O^′2 = +0.375 V) (Fig. 2a) assigned to Pd⁰(PPh₃)₃ and SPd⁰(PPh₃)₃ involved in the equilibrium of Scheme 1 [17]. When Br^– ions (2 equiv) were added to Pd⁰(PPh₃)₄, the two oxidation peaks O′₁ and O'₂ disappeared to give rise to a single oxidation peak O₁ (Fig. 2b), already obtained when PdBr₂(PPh₃)₂ was reduced in the presence of 2 equiv PPh₃ and which characterized the oxidation of the anionic Pd⁰ complexes [Pd⁰(PPh₃)_nBr]^– (n = 2 and 3) involved in the equilibrium in Eq. (6). When CO was bubbled into the solution of Pd⁰(PPh₃)₄ (2 mM) in THF containing 2 equiv Br^–, a new broad oxidation peak developed at O₄ (E^p_O4 = +0.245 V) (Fig. 2c), attesting the complexation of the [Pd⁰(PPh₃)_nBr]^– complexes by CO to form less easily oxidized neutral complexes Pd⁰(PPh₃)_n(CO) (n = 2 and 3) (Eq. (10)) – for the existence and characterization of Pd⁰(PPh₃)_n(CO) complexes, see [18]. The oxidation peak O₄ is similar to that observed when PdBr₂(PPh₃)₂ was reduced in the presence of CO and 2 equiv PPh₃ (Fig. 1b).

In another set of experiments, CO was first added to Pd⁰(PPh₃)₄. A broad oxidation peak O′₄ was observed at E^p_O^′4 = +0.485 V which characterized the oxidation of the Pd⁰(PPh₃)_n(CO) (n = 2 and 3) complexes involved in the equilibrium of Eq.(10). After addition of 2 equiv Br^–, the broad oxidation peak O′₄ disappeared and the only peak O₄ (E^p_O^′4=+0.241V) was detected. Since we have seen above that Br^– was released from anionic [Pd⁰(PPh₃)₂Br]^– in the presence of CO, the peak O₄ characterized the oxidation of Pd⁰(PPh₃)_n(CO) (n = 2 and 3) in the presence of Br^–. The observed potential shift from O′₄ to O₄ was therefore mostly due to the chemical reaction of Br^– with the transient Pd^I electrogenerated in the first electron oxidation of Pd⁰(PPh₃)_n(CO) (n = 2, 3) (EC mechanism). The potential of an irreversible peak depends on the rate of the chemical step that follows the first electron transfer. In the present case, the first electron transfer gives a cationic Pd^I complex involved in a chemical reaction with Br^– ions. This transient ‘Pd^IBr’ complex is more easily oxidized than the initial Pd⁰. This results in an overall bielectronic oxidation process occurring at the oxidation potential of Pd⁰, which gives eventually a Pd^II complex. For the mechanism of the electrochemical oxidation of Pd⁰ complexes, see [19]. The above results are summarized in Scheme 2.

Whatever the Pd⁰ precursor, Pd⁰(PPh₃)₄ and 2 equiv Br^– or the Pd⁰ complex generated by the electrochemical reduction of PdBr₂(PPh₃)₂ in the presence of 2 equiv PPh₃, the introduction of CO generates Pd⁰(PPh₃)_n(CO) (n = 2 and 3) complexes which are oxidized at O₄ (E^p_O4 = +0.245 V, Figs. 1b and 2c).

At room temperature and under atmospheric pressure, CO does not coordinate PdBr₂(PPh₃)₂ and does not reduce PdBr₂(PPh₃)₂. The only evolution of PdBr₂(PPh₃)₂ would then be its reduction by the anions ethoxide generated from EtOH in the cathodic process (Eq. (3)).

2.2 Reduction of PdBr₂(PPh₃)₂ by ethoxide ions in the absence and presence of PPh₃ (2 equiv)

The fate of PdBr₂(PPh₃)₂ in the presence of excess EtONa (used to mimic the effect of EtO^– electrogenerated at the cathode in the catalytic reaction) was monitored by cyclic voltammetry in THF. The oxidation potential of EtO^– was observed at +1.175 V in THF.

When 8 equiv EtO^– were added to a solution of PdBr₂(PPh₃)₂ (2 mM) in THF, its reduction peak R₁ disappeared and a succession of oxidation peaks were observed (Fig. 3), whereas the solution turned black. This suggests that PdBr₂(PPh₃)₂ was reduced by the ethoxide to a multitude of Pd⁰ complexes or clusters due to a lack of stabilizing ligands.

To observe stable Pd⁰ complex(es) detectable in cyclic voltammetry, the chemical reduction of PdBr₂(PPh₃)₂ by the ethoxide was performed in the presence of two equiv PPh₃. The reduction of PdBr₂(PPh₃)₂ by 2 equiv EtO^– was a slow reaction as attested by the slow decay of the reduction peak of PdBr₂(PPh₃)₂. A large excess EtO^– (8 equiv) was required to reduce PdBr₂(PPh₃)₂ quantitatively within a reasonable time (1 h 30). Fig. 4a exhibits the reduction peak of PdBr₂(PPh₃)₂ in the presence of 2 equiv PPh₃. When 8 equiv EtO^– were added, the reduction peak current of PdBr₂(PPh₃)₂ decreased (Fig. 4b, reaction time: 35 min) attesting a reaction of PdBr₂(PPh₃)₂ with EtO^–. An oxidation peak was observed on the reverse scan at a potential close to that of O₁, but with a higher current intensity than initially (compare Fig. 4a and b), attesting that a Pd⁰ complex was generated by the reaction of EtO^– with PdBr₂(PPh₃)₂. This Pd⁰ complex was indeed detected by its oxidation peak O₅ when the voltammetry was performed directly towards positive potentials, without scanning over peak R₁ (Fig. 4c). The complete reduction of PdBr₂(PPh₃)₂ by EtO^– was achieved within 1 h and a half and the resulting Pd⁰ was oxidized at O₅ (E^p_O5 = +0.090 V).

Due to the close vicinity of the oxidation potentials of O₁ and O₅, it is difficult to conclude from the cyclic voltammetry whether the Pd⁰ generated by the chemical reduction of PdBr₂(PPh₃)₂ was still ligated by Br^– or by EtO^–. Alcazar-Roman and Hartwig have reported kinetic evidence of the formation of anionic Pd⁰ complexes ligated by alkoxides [20]. To better characterize the Pd⁰ complex generated by the chemical reduction, EtO^– (8 equiv) was added to a solution of Pd⁰(PPh₃)₄ (2 mM) in THF. The two oxidation peaks O′₁ (+0.245 V) and O’₂ (+0.375 V) characteristic of Pd⁰(PPh₃)₃ and SPd⁰(PPh₃)₃ [17] were slightly shifted to more positive potentials O′₅ (+0.265 V) and O′₆ (+0.385 V) respectively. One would have expected to observe less positive oxidation potentials for the putative anionic [Pd⁰(PPh₃)_n(OEt)]^– species. However, the electrochemical oxidation of anionic [Pd⁰(PPh₃)_n(OEt)]^– species in the presence of EtO^– must be a complex process as attested by the irreversibility of the oxidation peaks. The formation of anionic [Pd⁰(PPh₃)_n(OEt)]^– (n = 2 and 3) complexes was better supported by ³¹P NMR experiments. Indeed, the broad ³¹P NMR signal at +3.23 ppm characteristic of Pd⁰(PPh₃)_n (n = 2 and 3) in equilibrium with PPh₃, generated from Pd⁰(PPh₃)₄ (7.9 mM) in THF (Scheme 1), was shifted to –3.0 ppm by addition of EtO^– (8 equiv). The signal of the free phosphine was not observed. This shows that EtO^– interfered in the equilibrium of Scheme 1, to form anionic species (Eq. (11)), as already established for halides or acetates [21].

The chemical reduction of PdBr₂(PPh₃)₂ requires two equivalents of EtO^– and the mechanism involves a classical β–hydride elimination [22], leading to a hydrido-Pd^II complex, whose HBr elimination by the basic ethoxide generates a Pd⁰ complex stabilized by excess EtO^– and PPh₃ (Scheme 3).

2.3 Reaction of ethoxide ions with Pd⁰(PPh₃)_n(CO) complexes

According to Scheme 2, Pd⁰(PPh₃)_n(CO) (n = 2 and 3) complexes were generated in the absence of EtO^–, whereas [Pd⁰(PPh₃)_n(OEt)]^– complexes were formed in the absence of CO (Scheme 3). Taking into account the fact that (i) CO is a better ligand than EtO^– for the electron rich Pd⁰(PPh₃)₂ complex and that (ii) CO is present in large amount in the very first step of the electrolysis whereas EtO^– is generated only when the electrolysis proceeds (Eq. (3)), we favour the formation of Pd⁰(PPh₃)₂(CO) as the main Pd⁰ complex in the very first step of the electrolysis (performed in the absence of PPh₃) rather than that of [Pd⁰(PPh₃)₂(OEt)]^–. It was thus of interest to investigate the reaction of EtO^– with Pd⁰(PPh₃)_n(CO) complexes, which was monitored in parallel by ³¹P NMR spectroscopy and cyclic voltammetry.

Pd⁰(PPh₃)_n(CO) (n = 2 and 3) complexes were first generated by bubbling CO into a solution of Pd⁰(PPh₃)₄ (8 mM) in THF and characterized by a broad ³¹P NMR signal at +7.43 ppm (Eq. (10)) and the broad oxidation peak O′₄ (E^p_O^′4 = +0.485 V) already mentioned above. Addition of EtO^– (8 equiv) generated a new broad signal at +4.15 ppm and a new oxidation peak at less positive potential O′₇ (E^p_O^′7 = +0.415 V) (Fig. 2d), i.e., new species. The reaction of strong nucleophiles such as alkoxydes RO^– with M⁰(CO)_n complexes is a known process that involves the nucleophilic attack of one CO ligand to form anionic alkoxycarbonyl [(CO)_n–1M⁰(COOR)]^– complexes [23,24]. In the present case, anionic Pd⁰ complexes [(PPh₃)_nPd⁰–COOEt]^– (n = 2 and 3) would be generated (Scheme 4). The reaction was monitored by IR spectroscopy. The CO vibration in Pd⁰(PPh₃)₃(CO) appeared at 2019 cm^–1 in THF, close to that reported for the solid (1955 cm^–1) [18]. When EtO^– was added to Pd⁰(PPh₃)₃(CO), a new absorption developed at 1650 cm^–1, characteristic of COOEt ligated to a palladium centre [11].

The bielectronic oxidation of [(PPh₃)_nPd⁰–COOEt]^– (n = 2 and 3) at O′₇ is expected to generate cationic Pd^II complexes [(PPh₃)_nPd^II–COOEt]⁺ (n = 2 and 3) (Scheme 4).

When 2 equiv of Br^– were added to the solution containing of [(PPh₃)_nPd⁰–COOEt]^– (n = 2 and 3), the oxidation peak O'₇ (E^p_O^′7=+0.415V) observed in Fig. 2d was replaced by a new one O₇ located at less positive potential (E^p_O7 = +0.215 V) (Fig. 2e). The easier oxidation of [(PPh₃)_nPd⁰–COOEt]^– (n = 2 and 3) in the presence of Br^– is most certainly featuring the chemical reaction of the bromide ion with the transient Pd^I formed in the first electron transfer (EC mechanism). After an overall bielectronic process, a neutral Pd^II complex BrPd^II–COOEt(PPh₃)₂ should be generated ultimately (Scheme 5). Related complexes ClPd–COOR(PPh₃)₂ are reported in literature. They have been synthesized by either the oxidative addition of Cl–COOR to Pd⁰(PPh₃)₄ [11,25,26].

The same oxidation peak O₇ was also observed when PdBr₂(PPh₃)₂ was reduced by EtO^– in the presence of PPh₃ (2 equiv) and CO (Fig. 4d).

2.4 Mechanism of the palladium-catalysed electrosynthesis of diethyl carbonate from carbon monoxide and ethanol

The above mechanistic investigations were developed in THF and in the presence of excess PPh₃, so that to have a chance to observe and characterize relatively stable Pd⁰ complexes and to be able to monitor their evolution in the presence of the reagents involved in the catalytic process. The electrosynthesis of diethyl carbonate proceeds in ethanol and in the absence of extra PPh₃ [3]. The catalyst PdBr₂(PPh₃)₂ was not tested in the electrosynthesis of diethyl carbonate but a related complex PdCl₂(PPh₃)₂. The latter was also successfully used in the electrocarbonylation of benzylamine to dibenzylurea [27]. A catalytic cycle for the electrosynthesis performed in ethanol is proposed in Scheme 6.

The chemical reduction of the catalytic precursor PdBr₂(PPh₃)₂ by the EtO^– generated in the cathodic process gives a Pd⁰ that is probably not ligated by neither Br^– nor EtO^–, but by CO. Indeed, the slow chemical reduction proceeds in the presence of CO and, as mentioned above, the concentration of CO in the electrolysis is constant (bubbling CO) and is always higher than that of EtO^– anions. The latter are generated at the cathode and their concentration is controlled by the current density as well as by their chemical consumption. Indeed, once EtO^– ions have reduced the PdBr₂(PPh₃)₂ to Pd⁰, they are also consumed in a chemical reaction with Pd⁰(PPh₃)₂(CO). This reaction gives the anionic complex [(PPh₃)₂Pd⁰–COOEt]^–. The key step of the anodic process is thus the overall bielectronic oxidation of [(PPh₃)₂Pd⁰–COOEt]^–. This oxidation is conducted in the presence of Br^– delivered by the precursor PdBr₂(PPh₃)₂. Consequently, it should generate the neutral complex BrPd^II–COOEt(PPh₃)₂ rather than the cationic complex [(PPh₃)₂Pd^II–COOEt(S)]⁺.

The last step of the catalytic cycle is a nucleophilic attack of the COOEt ligand of BrPd^II–COOEt(PPh₃)₂ by EtO^–, as detailed in Scheme 7. This reaction is supported by reported nucleophilic attack of methanol on related complexes, (AcO)Pd^II–COOMe(PPh₃)₂, which yields dimethyl carbonate only in the presence of a base [11].

The mechanism of the palladium-catalysed electrosynthesis of diethyl carbonate from carbon monoxide and ethanol described in Scheme 6 indicates that the anodic and cathodic reactions are interfering. As a result of the mechanistic investigation, the two processes may be expressed as:

In the presence of bromide ions, which occurs in the electrolysis because of the catalytic precursor PdBr₂(PPh₃)₂, the oxidation potential of Pd⁰(PPh₃)₂(CO) (E^p_O4 = +0.245 V) and that of [(PPh₃)₂Pd⁰–COOEt]^– (E^p_O7 = +0.215 V) are very close. If the nucleophilic attack of Pd⁰(PPh₃)₂(CO) by EtO^– is not fast enough (it is limited by the rate of production of EtO^– at the cathode) the oxidation of Pd⁰(PPh₃)₂(CO) will also occur leading back to PdBr₂(PPh₃)₂ and consequently would contribute to lower the faradic efficiency.

4.1 Electrochemical set-up for cyclic voltammetry

Electrochemical experiments were carried out in a three-electrode cell connected to a Schlenk line. The cell was thermostated at 25 °C. The steady working electrode consisted of a gold disk (i.d. 0.5 mm). The counter electrode was a platinum wire of ca 1-cm² apparent surface area. The reference was a saturated calomel electrode separated from the solution by a bridge filled with a solution of nBu₄NBF₄ (0.3 mol dm^–3) in 3 ml of THF. 12 ml of THF containing the same concentration of supporting electrolyte was poured into the cell followed by 17 mg (0.024 mmol) of PdBr₂(PPh₃)₂ and 12.5 mg (0.048 mmol) of PPh₃, when required.

In other experiments, 13 mg (0.19 mmol) of EtONa were added to the solution of PdBr₂(PPh₃)₂ in the absence or presence of bubbling CO.

The cyclic voltammetry was also performed on solutions of 28 mg (0.024 mmol) Pd⁰(PPh₃)₄ alone and then in the presence of 13 mg (0.19 mmol) of EtONa in the absence or presence of bubbling CO. When required, 15.5 mg (0.048 mmol) of nBu₄NBr were added.